On Fri, Oct 26, 2007 at 03:37:15PM +0930, Tim wrote: > On Thu, 2007-10-25 at 17:58 -0500, Isaac Serafino wrote: > > Is there any way to get and use Fedora without the RPM program or any > > RPM packages, for instance, using an alternative package manager, or > > compiling everything from the source? > > I'd have to wonder why you'd want to do that. You might as well do > Linux from scratch, or pick on of the other smaller distros which don't > use a package manager. One would like to do that because Fedora's innovation, engineering, and QA are very valuable, but RPM [or rather, its "coding in assembly" approach used in practice] is obsolete and not suitable in a networked world with distributed filesystems, virtualization, and lots of other configuration management headache multipliers. It is possible to hack most RPM specs to operate at a much higher level using macros, but the amount of work involved is such that converting to a different mechanism is probably just slightly more work. The whole discussion recently regarding multilib and the pain of creating separate *-libs subpackaging just makes me laugh/cry: with rPath Conary, the packaging system separates tagging, policy, and mechanism. Executables, shared libraries, and configuration files can all be treated differently *and* the policy is readily extensible / hookable. [Conary is not without its own warts, but what is?] There has been work done in Conary to extract tarballs and patches from SRPMS, http://wiki.rpath.com/wiki/Conary:RPM_Package_Recipe but I don't know of a mechanism for automatically converting a substantial fraction of spec files to Conary recipe format. In principle, it is possible to process the spec file to determine things like patch application order, as is done in quilt setup: http://utcc.utoronto.ca/~cks/space/blog/linux/PatchingRPMsWithQuilt "Vanilla" rpm spec scripts that use %configure, %makeinstall, etc., should be rather trivial to convert. Regards, Bill Rugolsky
